HP 8590A spectrum analyzer repair

[feedback.loop]

Repair of 8590A analyzer, which turned out to have bad mixer diodes in the first converter:



I wonder if anyone has a better service manual. The one from Keysight seems to be incomplete and has many corrections in handwriting. Some adjustment procedures are not clear at all.
I would appreciate a better one.
Thanks.

[SaabFAN]

Nice! I always thought that a fault in the first Mixer was the death-sentence for these Spectrum Analyzers.
Good to be proven wrong

[MadTux]

Interesting video! Didn't knew, those 8590A are build that noob friendly. I would have expected some wirebonded magic instead of discrete, soldered glass diodes.

Good work

Some pics of the 2-22GHz YTF-Mixer assembly of one of my 8566 spectrum analyzers. Mixer diode is under a extremely fine gold wire mesh in the assembly in the center, close to the blue blue blob.

 The YIG filter is also in there, 3 YIG balls mounted onto ceramic rods for fine adjustment. In the back you can see the large magnet coil used for tuning those YIG filters. If you decide to take it apart, you need to mark the rotation of those magnet coils relative to the case, as the magnetic poles are slightly slanted, so they can be aligned to get the most homogeneous magnetic field between them.

Fry that and you're done. Instrument works perfectly now, btw. YTF tuning current vs frequency was misaligned, which caused the amplitude to drop at higher frequencies.

Anyone some pics of the 8568 1th mixer? Haven't had that one open, but would be quite curious, since that is my everyday use SA.

[AF6LJ]

Nicely Done...
I think I would have bought the diodes with 15V breakdown.
Still that is great to have a working analyzer for such a reasonable price. 

[feedback.loop]

Nicely Done...
I think I would have bought the diodes with 15V breakdown.
Still that is great to have a working analyzer for such a reasonable price. 

It seems that 282x family (15V breakdown) has higher noise. Here is a quote from the datasheet:
"The HSMS-281x family is a hybrid Schottky (as is the HSMS-280x), offering lower 1/f or flicker noise than the HSMS-282x family."

[MadTux]

......offering lower 1/f or flicker noise than the HSMS-282x family."

Wouldn't worry about that. 859X goes down to only 10kHz and doesn't have narrow RBW, so that the low frequency range, in which the 1/f (pink noise) is relevant, has high noise by the LO anyway.

[xrunner]

Excellent work and video. I have a HP 8595E myself.

[chris_leyson]

Nice repair, and a mini teardown as well, thank you 

It's always nice to see teardowns for analyzers at this sort of frequency range, at least some of the RF stage is repairable. I like the way HP put the low pass filters into the first mixer box, saves on external filters and connectors, and that's the first time I've seen ferrite used to absorb harmonics. Was surprised to see a matched pair of diodes I was expecting a to see a bridge for some reason. Nice design 

[feedback.loop]

Followup - experiments with different diodes in the mixer:

[tggzzz]

Followup - experiments with different diodes in the mixer:

Is there a TL;DW summary somewhere? I'm not going to spend 40minutes of my (remaining) life finding out whether or not you have done something that would be useful to me. If the results are good then I will watch.

[DimitriP]

Followup - experiments with different diodes in the mixer:

Is there a TL;DW summary somewhere? I'm not going to spend 40minutes of my (remaining) life finding out whether or not you have done something that would be useful to me. If the results are good then I will watch.

Heeeey...it takes one click on the Youtube logo to see the summary you are seeking ....

[tggzzz]

Followup - experiments with different diodes in the mixer:

Is there a TL;DW summary somewhere? I'm not going to spend 40minutes of my (remaining) life finding out whether or not you have done something that would be useful to me. If the results are good then I will watch.

Heeeey...it takes one click on the Youtube logo to see the summary you are seeking ....

I see a link to this thread, a service manual and some datasheets. From that I can infer "he did some work", but cannot infer the results.

[feedback.loop]

Is there a TL;DW summary somewhere? I'm not going to spend 40minutes of my (remaining) life finding out whether or not you have done something that would be useful to me. If the results are good then I will watch.

I see a link to this thread, a service manual and some datasheets. From that I can infer "he did some work", but cannot infer the results.

You must be referring to the first video. I believe that the summary there describes what was done:
"A replacement was found and installed. The analyzer seems to be back in business. Perhaps, it will need some adjustments though."

By the way, what is "TL;DW summary"?

If the results are good then I will watch.

Would you watch a failed repair attempt?

[DimitriP]

There are plenty of timewaster videos out there to complain about, this is not one of them.
Not sure what the issue is, or why it is with this one in particular.

If it were a video about comparing soldering tips, or DMM probes I'd understand

[PA4TIM]

Hi feedback.loop. I like your channel (and your accent  ) One of the better ones I know. Always nice to see how others think and do when it is about repairing measurement gear. 

[nidlaX]

Hey feedback.loop, keep making videos! I've watched a few of your other ones such as the Keithley electrometer repair and the Agilent 66309D repair. Good work!

As a bit of feedback (), I suggest you record the whole teardown and play back the less interesting bits at increased speed. This way, someone else who is working on the same instrument can see the entire teardown procedure, while you can avoid boring the viewer.

[Shock]

Been watching your channel since it started, great videos.

[tggzzz]

"A replacement was found and installed. The analyzer seems to be back in business. Perhaps, it will need some adjustments though."

I'd like to know what he means by "business". For some ebay sellers it is sufficient that "it turns on and the trace wiggles"

By the way, what is "TL;DW summary"?

Too long; didn't watch

Would you watch a failed repair attempt?

Yes, if I had reason to believe I could learn something from it.

If there's no visible reason why I might learn, well there are many many many other things where I can see I will learn something. Guess what will get my attention.

Always remember the "elevator pitch" concept - and go on to back that up with solid information.

[Shock]

I send an email to the youtube standards committee, they will investigate this matter thoroughly. 

[AF6LJ]

Just a point of observation; every SA I have used that had a first LO out had it terminated in a 50 ohm load when it was not in use.

[feedback.loop]

I wonder if anyone has a newer version of firmware. My unit has 9.1.87. According to service manual, there are at least three more versions: 31.7.87, 29.7.88 and 14.10.88.

[DimitriP]

Firmware jungle....before you conclude the wrong thing, ask over at  https://groups.yahoo.com/neo/groups/hp_agilent_equipment/info.

[azer]

Firmware can be had from here ftp://ftp.bluefeathertech.com/electronics/testgear/HP_Agil_Keysight/firmware/8590A/
The version number is just the revision date, 14.10.88 is the latest one before they changed to 8590E series.

[feedback.loop]

It turned out that there was another issue. I found that the input attenuator Wavetek 0955-0329 was defective in a strange way: the 20 dB pad in it attenuated lower frequencies only by 15 dB, but worked fine at higher frequencies. I wonder if it is possible to take it apart without destroying it. I found a replacement for now, but I would love to fix it. The replacement is Wavetek 0955-0453, which I borrowed from another non-working HP 8590A unit. It is slightly longer physically, but has the same mounting holes, and otherwise seems to be identical.

Part 3 video:

[Shock]

Well done, nice detective work.

[AF6LJ]

Good to see it is working Great Job.

[feedback.loop]

I made a lot of progress with the parts unit - the one with 75 Ohm input, which I showed in the part 3 video. There was no display, but bright vertical columns all over the screen. By swapping boards I found that the video board, the main board and the analog interface board were bad.

By the way, I forgot to mention in the video, and someone asked on YouTube - how come I could borrow the attenuator from it and put it into the 50 Ohm unit. Answer: The units are almost identical. They have only two small differences: the 75 Ohm unit has 75-to-50 matching pad right at the input connector, and the 3rd converters differ slightly - they have different impedance of the calibration output, which goes to the front panel.

The main board was easy to fix. I had a closer look at it and found that one of the four EPROMS had corrosion on the pins so that three pins broke off. But, strangely, the rest of the board, and even the socket looked clean. This EPROM was right next to the fan, so, perhaps, something corrosive got in through the fan. I replaced the chip (programmed the right part of the firmware into it), and the main board started right up!

On the video board all four TMS4416 memory chips were dead. After replacing them the video board woks fine.

On the analog interface board four out of five 1826-1386 DACs were dead. I figured that they are AD7541, and replaced with AD7541AKN versions, which have the best parameters (except, perhaps, some military versions). Now I can put the board into the working analyzer, and it works the same (after self calibration).

So now it seems that I have a working analyzer, but I borrowed the input attenuator from it.

Question: Does anybody know how to take apart these Wavetek attenuators (0955-0329 or similar)? Is it possible without destroying them? Is it worth the effort - meaning if they are repairable? I would appreciate your suggestions. Thanks!

[bitseeker]

Wow, that's great progress on the donor unit!

Regarding the 75-ohm input, does that mean you can easily convert it to 50-ohms or is there more involved?

[feedback.loop]

Regarding the 75-ohm input, does that mean you can easily convert it to 50-ohms or is there more involved?

As I said, I believe there are two differences in 75 ohm units:
1. matching pad right at the input connector (and BNC connector instead of N)
2. different version of the 3rd converter board so that the cal output is 75 ohm
We can add one more little thing, which is the model name on the front panel. 75 ohm units are marked as "option 01" right under the 8590A model name.

The first must be easy to change except the size of the hole in the front panel. The second thing I am not sure, but I would imagine that it shouldn't be very hard. The hardest part is to find schematics for both to know for sure what to change.

[feedback.loop]

Regarding the 75-ohm input, does that mean you can easily convert it to 50-ohms or is there more involved?

As I said, I believe there are two differences in 75 ohm units:
1. matching pad right at the input connector (and BNC connector instead of N)
2. different version of the 3rd converter board so that the cal output is 75 ohm
We can add one more little thing, which is the model name on the front panel. 75 ohm units are marked as "option 01" right under the 8590A model name.

The first must be easy to change except the size of the hole in the front panel. The second thing I am not sure, but I would imagine that it shouldn't be very hard. The hardest part is to find schematics for both to know for sure what to change.

Here are photos of both 50 ohm and 75 ohm versions of the 3rd converter

[feedback.loop]

As I said, I believe there are two differences in 75 ohm units:
1. matching pad right at the input connector (and BNC connector instead of N)
2. different version of the 3rd converter board so that the cal output is 75 ohm
We can add one more little thing, which is the model name on the front panel. 75 ohm units are marked as "option 01" right under the 8590A model name.

The first must be easy to change except the size of the hole in the front panel. The second thing I am not sure, but I would imagine that it shouldn't be very hard. The hardest part is to find schematics for both to know for sure what to change.

Oops, I am sorry. I don't know what I was thinking. Both versions have BNC input connector. So the front panel only differs by that "option 001" marking below the model name.

[vtp]

Question: Does anybody know how to take apart these Wavetek attenuators (0955-0329 or similar)? Is it possible without destroying them? Is it worth the effort - meaning if they are repairable? I would appreciate your suggestions.

I have repaired one similar HP attenuator but in my case it was JFW Industries model. It had a screw-on top plate. Your Wavetek seem to have different casing construction though feedtrough capacitors look to be right in the same position as in JFW and both had also 10, 20 and 30 dB pads where 10 dB pad was selected by default.

The JFW attenuator was build with 1206 resistors and through-hole microwave relays on rogers or similar PCB substrate. The only problem I had was that the PCB (and ground end of 1206 resistor pads) were soldered to the casing so it took quite a lot of heat to get blown resistors off the PCB. I suspect that they used ordinary thick film resistors but I put in thin films as replacement.

Eventually, I got the job done and used 8753E VNA to check the attenuator and it became better or equal to another similar unit. Insertion loss for these seem to be about 2dB or so.

If you can figure out how to open that Wavetek attenuator it should be repairable if it is anything like the JFW attenuator in construction and then certainly well worth the effort.

[azer]

I believe option H07 was for N connector input.

[k1mgy]

The main board was easy to fix. I had a closer look at it and found that one of the four EPROMS had corrosion on the pins so that three pins broke off. But, strangely, the rest of the board, and even the socket looked clean. This EPROM was right next to the fan, so, perhaps, something corrosive got in through the fan. I replaced the chip (programmed the right part of the firmware into it), and the main board started right up!

Would you recommend a thin removable conforming coating (spray) to protect this area?  Might consider it for my 8590a.

Fantastic work on that 75Ohm unit.  Hope to see a video on this one.

[Armadillo]

It turned out that there was another issue. I found that the input attenuator Wavetek 0955-0329 was defective in a strange way: the 20 dB pad in it attenuated lower frequencies only by 15 dB, but worked fine at higher frequencies. I wonder if it is possible to take it apart without destroying it. I found a replacement for now, but I would love to fix it. The replacement is Wavetek 0955-0453, which I borrowed from another non-working HP 8590A unit. It is slightly longer physically, but has the same mounting holes, and otherwise seems to be identical.

Very helpful and educational. Since you found the replacement, its time to do a teardown of the Wavetek 0955-0329 in your next video. I am looking forward to it. thanks.

[feedback.loop]

It turned out that there was another issue. I found that the input attenuator Wavetek 0955-0329 was defective in a strange way: the 20 dB pad in it attenuated lower frequencies only by 15 dB, but worked fine at higher frequencies. I wonder if it is possible to take it apart without destroying it. I found a replacement for now, but I would love to fix it. The replacement is Wavetek 0955-0453, which I borrowed from another non-working HP 8590A unit. It is slightly longer physically, but has the same mounting holes, and otherwise seems to be identical.

Very helpful and educational. Since you found the replacement, its time to do a teardown of the Wavetek 0955-0329 in your next video. I am looking forward to it. thanks.

That is a possibility. I was busy with other things lately, but I do plan to make such videos, and this project is in the list.

[Armadillo]

Followup - experiments with different diodes in the mixer:

Is there a TL;DW summary somewhere? I'm not going to spend 40minutes of my (remaining) life finding out whether or not you have done something that would be useful to me. If the results are good then I will watch.

I chanced open up the first mixer of HP8591E but it was using the model 08591-60113, as you can see in the photo, its a more elaborate in construction than the one you video. The schottky mixer diode as marked is the C2E, I think is the HSM276S. But note the reverse breakdown at Vr=0.5V and Ir > 100uA will fry the diodes, so static caution need to be observed. The HSM276S is a balanced mixer diode [not detector diode] so it should not be probed with a multi-meter.
But I have no way of knowing whether it would be better then Avago HSMS series you tried.
I am looking forward to your next video.

[feedback.loop]

I chanced open up the first mixer of HP8591E but it was using the model 08591-60113, as you can see in the photo, its a more elaborate in construction than the one you video. The schottky mixer diode as marked is the C2E, I think is the HSM276S. But note the reverse breakdown at Vr=0.5V and Ir > 100uA will fry the diodes, so static caution need to be observed. The HSM276S is a balanced mixer diode [not detector diode] so it should not be probed with a multi-meter.
But I have no way of knowing whether it would be better then Avago HSMS series you tried.
I am looking forward to your next video.

How did you figure out that the devices marked C2E are HSM276S? I believe they are HSMS-2822. Have a look at the data sheet. Mine were marked C2X, and the last character represents date code.

[Armadillo]

How did you figure out that the devices marked C2E are HSM276S? I believe they are HSMS-2822. Have a look at the data sheet. Mine were marked C2X, and the last character represents date code.

The HSM276S had me confused for a while since it is marked C2. With the extra date code digit emh!, I am sure you are right, its the HSMS-2822.
Thanks for correction. 

[petercut62]

Hello, Keysight has given the method to switch 50 to 75 or 75 to 50

May be useful for some of us

Found here : https://community.keysight.com/thread/5431
----------
When using an external matching pad or network, the analyzer does not know that it is connected, so you have to compensate for that 5-6 dB front end loss for sure.  If you plan to convert the analyzer, you can find the details below.  Since your analyzer is very old, you may not be able to get the conversion parts needed, not sure.

Agilent 8590B and HP 8591A Conversion for 50 or 75 ohm Operation
________________________________________
Parts listed for the impedance input required. Order one of each.     

75 Ohm      Description      50 Ohm     
0590-1251     NUT HEX 15/32-32      NONE     
08590-20182     CBL AY-RF INPUT     08590-20157     
08590-60179     RF INPUT      1250-2191     
NONE      ADPT M N - F BNC     1250-0780     
6960-0150      PLUG-HOLE      6960-0148     
08590-60028      Cable, Cal Out      8120-5052     
5062-6452 (75ohm)     CABLE, CAL Out-RF In     8120-2682 (50ohm)     
0955-0453     A24A4 ATTENUATOR, TG OUTPUT     0955-0453     
08590-20179     J2 MATCHING PAD     NONE     


Agilent 8590B Dress Panel
     
75 Ohm      Description      50 Ohm     
08590-00039     HARD KEYPAD     08590-00038     
08590-00046     RUBBER KEYPAD     08590-00045     

HP 8591A Dress Panel
     
75 Ohm     Description     50 Ohm     
08591-00002     HARD KEYPAD     08591-00001     
08591-00014     RUBBER KEYPAD     08591-00013
     
Common to the HP8590B and the Agilent 8591A conversions.
To convert a 50 ohm 3RD CONVERTER, p/n 08590-60192, 08593-60021 or 08591-60081 to 75 ohm, order the parts listed in the 75 ohm column and replace the parts listed in the 50 ohm column.
To convert a 75 ohm 3RD CONVERTER, p/n 08590-60193 or 08590-60344 to 50 ohm, order the parts listed in the 50 ohm column and replace the parts listed in the 75 ohm column.

     
75 Ohm      Description     50 Ohm     
0698-7239, 1.33 KOHM     A9R20     0698-7237, 1.1 KOHM     
0698-7216, 147 OHM     A9R21     0698-7212, 100 OHM     
0698-7216, 147 OHM     A9R23     0698-7212, 100 OHM     
9135-0076, 39 NH     A9L2     9135-0070, 24 NH     
The RF input cable (W10) that connects to the rear of the RF Input connector remains the same for both configurations since it attaches to the rear of the Minimum Loss Adapter or the N connector.
Refer to "STOR PWR ON UNITS" in Chapter 13 (Service Softkey Descriptions) in the service manual for setting the Impedance and Power-On units of the analyzer to 50 or 75 ohm input.

Special considerations for 75 ohm spectrum analyzers.
If the final configuration is 75 ohms (Opt. 001), use only the 75 ohm cable (5062-6452) for calibration and use only 75 ohm cables for signal input. Using a 50 ohm cable can damage the 75 ohm input connector which is the Minimum Loss Adapter (MLA). The replacement cost of the MLA is currently $220. The MLA is not field repairable.
The recommended 75 ohm F to 75 ohm BNC connector p/n is 1250-0076, using Radio Shack type connectors WILL eventually damage the input connector because it is a 50 ohm BNC connector.
If converting an Option 011 (75 ohm TG) to 50 ohms, remove the MLA (TG Output Adapter, 50 to 75 ohm) and re- place with the type N (f), p/n 1250-2191, connector.

Special Considerations for the Agilent 8590B with a Count Lock, Option 013
If the 8590B has a Count Lock Option (Option 013), the output attenuator (0955-0453) will have to be added to the TG output or the TG Cal will fail. The reason is the analyzer will now ID as an Agilent 8591A, so when the Cal TG test routine is run the firmware will attempt to change the TG attenuator and test the results.

Regards -

[Bjoern]

1.   I tried to measure the signal path from the RF to the IF with a 500Mhz scope which I borrowed but I cannot see any sensible signal between the output and the input of the different stages.
Of course at 2.4 Ghz I would expect issues with the scope shorting the signal due to capacitance but at 300 Mhz I expected it to work. When I disconnect the wires from the SA I see on the screen that the signals disappear hence there must be a signal inside the cable its just that my scope doesn’t seem to pick it up.
Do you have a recommendation how I could check the signals, I do not have another spectrum analyser. I tried a small hand made coil (contact less) but didn’t help either.
Is it impossible to measure in your opinion ?
2.   I see a nice signal on the SA at a SPAN of 100khz and above.
The moment I tur the nob below 100Khz the signal becomes useless.
Can you recommend where to search ?

https://www.eevblog.com/forum/repair/hp8590l-spectrum-analyzer-calibration-signal-issues/new/?topicseen#new

Highly appreciated
Bjoern

[crasyboye]

Thank you , I repaired my HP8590A after watching your video, thank you very much.

[crasyboye]

The pictures bfore repairing and now.

[Johnny10]

feddbackloop,

What does the Cal Data look like in the Analyzer that you replaced the DAC's in?

[feedback.loop]

feddbackloop,

What does the Cal Data look like in the Analyzer that you replaced the DAC's in?

It has been a while. I found this photo.

[ndemarco]

The Avago/Broadcom diodes are discontinued, but available on eBay and Amazon. I just bought a lifetime supply (200) for $11 shipped.

FeedbackLoop - Thanks for the great video. You helped me bring my 8591E back to life. Now on to the 8563EC.